CENTRO DE TECNOLOGIA E GEOCIÊNCIAS DEPARTAMENTO DE OCEANOGRAFIA
PROGRAMA DE PÓS-GRADUAÇÃO EM OCEANOGRAFIA
Thiago Luiz do Vale Silva
Ocean-atmosphere processes and easterly waves in the generation and development of extreme events in the Eastern Northeast of Brazil
Recife
2018
Ocean-atmosphere processes and easterly waves in the generation and development of extreme events in the Eastern Northeast of Brazil
Thesis presented to the postgraduate program in oceanography of the Departmento of Oceanog-raphy of the Federal University of Pernambuco, as a requirement to obtain a doctorate degree in oceanography.
Main area: Abiotic Oceanography.
Supervisor: Prof. Dra. Dóris Regina Aires Veleda
Recife
2018
Catalogação na fonte
Bibliotecária Maria Luiza de Moura Ferreira, CRB-4 / 1469 S586o Silva,Thiago Luiz do Vale.
Ocean-atmosphere processes and easterly waves in the generation and
development of extreme events in the Eastern Northeast of Brazil/Thiago Luiz do Vale - 2018.
102folhas, il.;tab., abr., sigl. e simb.
Orientadora: Profª.Drª. Dóris Regina Aires Veleda.
Tese (Doutorado) – Universidade Federal de Pernambuco. CTG.Programa de Pós-graduação em Oceanografia, 2018.
Inclui Referências.
Texto em inglês.
1.Oceanografia.2. Camada limite atmosférica marinha.3.Camada de inversão térmica dos ventos Alísios.4. Piscinas quentes do Atlântico tropical sul (SAWP). Veleda,Dóris Regina Aires(Orientadora).II. Título.
UFPE
551.46CDD (22. ed.) BCTG/2018-271
Ocean-atmosphere processes and easterly waves in the generation and development of extreme events in the Eastern Northeast of Brazil
Thesis presented to the postgraduate program in oceanography of the Departmento of Oceanog-raphy of the Federal University of Pernambuco, as a requirement to obtain a doctorate degree in oceanography.
Work approved. Recife, 26 March 2018:
Dóris Regina Aires Veleda Supervisor
Carmen Medeiros Limongi
Rita Marques Alves
Washington Luiz Félix Correia Filho
Gbekpo Aubains Hounsou-gbo
Recife
2018
I thank to my family who gave me all kinds of educational, financial and affectionate supports.
I thank to my parents and grandmother for moral, disciplinary and human constitution that I have.
To my wife Elis Guimarães for all your love and companionship at all times, whether bad or good.
To my advisor Profa. Dr. Dóris Veleda, for the teachings and guidance during doctorate. And thank you for the opportunity and trust.
To the Prof. Dr. Moacyr Araújo for his guidance.
To the Prof. Dr. Alexandre Costa (CER) for the trust and opportunity in the participation of the projects and in the computational supports.
To HPC4E - High Performace Computing for Energy Program project for the opportunity to expand my vision as a researcher.
To the friend Dr. Pedro Tyançã for the talks, advice and for offering his computer "state of the art" to carry out the experiences of this thesis.
To my colleagues, who in bad times, raised my self-esteem. To the friends, who in bad times always raised my self-esteem. To the teachers from DOCEAN.
To the colleagues from DOCEAN.
Patrice Oliveira, APAC’s manager, who helped me during the doctoral period.
To GMMC / APAC colleagues in the days of struggle, of joy, of sadness, each in his own way, but all together and mixed together.
The coupled ocean-atmosphere wave and sediments transport (COAWST) model was used to simulate periods with intense rainfalls events at the eastern Northeast Brazil (ENEB). The simulations aimed to investigate the ocean-atmospheric interaction under Easterlies Waves Disturbances (EWD) conditions, which lead to intense rainfalls events over the ENEB. In June 2010 the sea surface temperature (SST) was warmer than 28.5◦C, in the western tropical South Atlantic (WTSA), with persistent anomalies above 1◦C, characterizing a southern Atlantic warm pool (SAWP). The sensible and latent heat fluxes acted to moisten the lower troposphere and affect the height of trade winds inversion layer (TWIL). The meteorological system, that occurred in low to medium levels during the period, favored the weakening, even to a break of the TWIL. These atmospheric disturbances were associated to convergence, cyclonic vorticity and upward water vapor motion to the medium troposphere levels. The disturbances when reached the ENEB coast, favored the convection and the intense rainfall over the region. The vertical structure of Marine Atmospheric Boundary Layer (MABL) was a key factor to triggering the occurrence of these synoptic activities with a deep convection. The sea-air interaction is important for understanding the climatic variability, due to heat, momentum and moisture exchange. Therefore the SAWP region is an important region to increasing meteorological systems over ENEB leading the variability and intense episodes of rainfall. In this work it was performed experiments with the Weather Research and Forecasting (WRF) and Regional Oceanic Model System (ROMS) in a two-way coupled mode. The goal was verify the improvements of coupled models in forecasting and characterizing the interactions between the ocean-atmosphere dynamics under different SAWP conditions which led intense rainfalls at the ENEB. The results evidenced the importance of the application of ocean-atmosphere coupled models in forecasting improvement compared to only atmospheric models, also the SAWP presented direct influence in the ENEB intense rainfall.
Keywords: Marine atmospheric boundary layer. Trade winds inversion layers. Southern Atlantic warm pool (SAWP).
O modelo coupled ocean-atmosphere wave and sediments transport (COAWST) foi utilizado para simular períodos de eventos intensos de precipitação sobre o leste do Nordeste Brasileiro (ENEB). As simulações tiveram como objetivo investigar as interações oceânicas e atmosféricas em condições de Distúrbios Ondulatórios de Leste (DOLs) os quais ocasionaram precipitações intensas no ENEB. Em junho de 2010 a temperatura superficial do mar (SST) registrada esteve acima de 28.5◦C, representando uma warm pool no Atlântico tropical sul (SAWP). Os calores sensíveis e latente agiram para umidificar anomalamente a baixa troposfera e afetar a altura da base das inversões térmicas dos Alísios (TWIL). O sistema meteorológico, o qual ocorreu de baixos a médios níveis durante o período, favoreceu o enfraquecimento, até a possível quebra da TWIL. Os DOLs são associadas a convergência, vorticidade ciclônica e vapor de água ascendente para a média troposfera. Os distúrbios ao chegar sobre a costa do ENEB, favoreceu convecção e precipitações extremas sobre a região. A estrutura vertical da Camada Limite Atmosférica Marinha (MABL) foi um fator chave para o gatilho da convecção profunda sobre a região. O entendimento sobre a interação oceano-atmosfera é importante para entender as variabilidades climáticas, devido aos fluxos de calor, momento e umidade. Portanto a região da SAWP é uma importante região para o aumento na intensidade dos sistemas meteorológicos sobre o ENEB gerando variabilidades climáticas e intensos episódios de chuvas. Neste trabalho foram feitas simulações com os modelos Weather Research and Forecasting (WRF) e o Regional Oceanic Model System (ROMS) em um modo acoplado em two-way. O objetivo foi avaliar o aperfeiçoamento das previsões e a analizar as características dinâmicas das interações entre oceano e atmosfera sobre diferentes condições na SAWP as quais levaram a intensas precipitações no ENEB. Os resultados mostraram um melhoramento das resoluções numéricas das previsões com modelos acoplados, ainda pôde observar influências diretas da SAWP nos eventos intensos de chuva.
Palavras-chave: Camada limite atmosférica marinha. Camada de inversão térmica dos ventos Alísios. Piscinas quentes do Atlântico tropical sul (SAWP).
Figure 1 – Northestern Brazil topography. Near coastal areas elevation in 0 - 200m
height, meanwhile central areas varying from 200 - 1200m in height. . . 25
Figure 2 – Brazilian Northeast rainfall climatology zones. I - Is the Southern Northeast Brazil, with maximum precipitations between November and February; II -Northern Northeast Brazil, with maximum precipitation between February and May; III - Eastern Northeast Brazil, maximum precipitation between May and August. . . 26
Figure 3 – Schematic representation of mean zonal and meridional currents are indicated: The upper ocean northward NBUC flow (red solid line), and the deepest southward transport by the DWBC (blue dashed line). The sSEC bifurcation at 100m depth (solid line); between 200m and 500m depth (dashed line); between 500m and 1200m depth (pointed line), according with (STRAMMA; ENGLAND, 1999). . . 27
Figure 4 – Trade Wind inversion layer Structure . . . 29
Figure 5 – Satellite image. Day 08 January 2018 at 0000UTC. . . 29
Figure 6 – A flow chart of the coupled code forming a single-executable modeling system running in concurrent mode. . . 34
Figure 7 – A flow chart of the coupled code forming a single-executable modeling system running in concurrent mode. . . 36
Figure 8 – Flowchart of parameterization information exchange in WRF. . . 37
Figure 9 – Studied Domain. Squared with lines is the SAWP region. Squared with crosses hatched is the Gradient regions. Greens dots is the PIRATA buoy. Red dots is the INMET PCD. Yellow dots is APAC pluviometers. Magenta dots is AESA pluviometers. Blue dots is EMPARN pluviometers. Shaded color is the Earth elevations (bathymetry and topography). . . 40
Figure 10 – Time Series of air temperature in 2m for a) Garanhuns; b) Recife and c) PI-RATA Buoy (8◦S e 30◦W); and Bias (simulated - measured) for d) Garanhuns e) Recife and f) measured at PIRATA buoy (8◦S e 30◦W) . . . 49
Figure 11 – Same as Figure 11, but for Relative Humidity . . . 50
Figure 12 – Same as Figure 10, but for uwindspeed . . . 51
Figure 13 – Same as Figure 10, but for vwindspeed . . . 52
Figure 14 – Same as Figure 10, but for wind speed . . . 53
Figure 15 – Sea surface temperature and surface currents simulated with the WRF-ROMS 2 experiment, a) in 10 June 2010 at 0000 UTC and b) 25 June 2010 at 0000 UTC. . . 58
(white lines) and a) experiment WRF-ROMS 1, b) experiment WRF-ROMS 2, c) experiment WRF-ROMS 3 (black lines and color shading), from 10 to 25 June 2010. . . 59 Figure 17 – Longitudinal profile, period averaged from 10 - 25 June 2010, of the
atmo-spheric water vapor (upper) and ocean temperature (bottom) in the exper-iments a) WRF-ROMS 1, b) WRF-ROMS 2c) WRF-ROMS 3, difference between (d) WRF-ROMS 1 and WRF 1, (e) WRF-ROMS 2 and WRF 2, (f) WRF-ROMS 3 and WRF 3and (g) WRF-ROMS 2 and WRF-ROMS 3. . . . 62 Figure 18 – Streamlines and vorticity from 12 June 2010 at 0000 UTC to 15 June 2010 at
0000 UTC for: 850hPa (a, b, c, d); 750hPa (e, f, g, h); and mid-levels water vapor content (i, j, k, l). . . 63 Figure 19 – Streamlines and vorticity from 16 June 2010 at 0000 UTC to 19 June 2010 at
0000 UTC for: 850hPa (a, b, c, d); 750hPa (e, f, g, h); and mid-levels water vapour content (i, j, k, l). . . 64 Figure 20 – Streamlines and vorticity from 21 June 2010 at 0000 UTC to 24 June 2010 at
0000 UTC for: 850hPa (a, b, c, d); 750hPa (e, f, g, h); and mid-levels water vapour content (i, j, k, l). . . 65 Figure 21 – Hovmöller of water vapor in medium levels, with longitude fixed at 35◦W,
for the experiments (a - upper panel) ROMS 1(a - middle panel) ROMS 2 and (a - bottom panel) ROMS 3. Difference between WRF-ROMS 1 and WRF 1 (b - upper panel), WRF-WRF-ROMS 2 and WRF 2 (b - middle panel), WRF-ROMS 3 and WRF 3 (b - bottom panel). . . 67 Figure 22 – Longitudinal profile of atmospheric water vapor (upper panel), period 10-25
June 2010 average removed, and ocean temperature (bottom panel) from (a) 12 June 2010 at 0000 UTC, (b) 13 June 2010 at 0000 UTC, (c) 14 June 2010 at 0000 UTC and (d) 15 June at 0000 UTC, Isolines of 0.012kg / kg (upper panel, black line) and oceanic temperature (bottom panel, black line) with emphasis on the isotherm of 20◦C and 28.5◦C. . . 69 Figure 23 – Longitudinal profile of atmospheric water vapor (upper panel), period 10-25
June 2010 average removed, and ocean temperature (bottom panel) from (a) 16 June 2010 at 0000 UTC, (b) 17 June 2010 at 0000 UTC, (c) 18 June 2010 at 0000 UTC and (d) 19 June at 0000 UTC, Isolines of 0.012kg / kg (upper panel, black line) and oceanic temperature (bottom panel, black line) with emphasis on the isotherm of 20◦C and 28.5◦C. . . 70
21 June 2010 at 0000 UTC, (b) 22 June 2010 at 0000 UTC, (c) 23 June 2010 at 0000 UTC and (d) 24 June at 0000 UTC, Isolines of 0.012kg / kg (upper panel, black line) and oceanic temperature (bottom panel, black line) with emphasis on the isotherm of 20◦C and 28.5◦C. . . 71 Figure 25 – Cumulative rainfall in the period from 10 to 16 June 2010 simulated with the
experiments: (a) WRF-ROMS 1, (b) WRF 1, (c) WRF-ROMS 2 (d) WRF 2, (e) WRF-ROMS 3 e (f) WRF 3 . . . 72 Figure 26 – Cumulative rainfall in the period from 16 to 19 June 2010 simulated with the
experiments: (a) WRF-ROMS 1, (b) WRF 1, (c) WRF-ROMS 2 (d) WRF 2, (e) WRF-ROMS 3 e (f) WRF 3.WRF-ROMS 3 e (f) WRF 3 . . . 73 Figure 27 – Cumulative rainfall in the period from 20 to 25 June simulated with the
experiments: (a) WRF-ROMS 1, (b) WRF 1, (c) WRF-ROMS 2 (d) WRF 2, (e) WRF-ROMS 3 e (f) WRF 3 . . . 74 Figure 28 – Averaged SST for 10 to 25 June with a) 2010 SST and b) 2012 SST boundary
conditions. Black contours is the 28.5◦C isotherm. . . 75 Figure 29 – Streamlines and wind speed in 850 hPa from day 16, 17 and 18 June 2010 in
WRF-ROMS using 2010 SST a, b, c and 2012 SST boundary condition d, e, f. 76 Figure 30 – Time series of SST-T2m averaged in SAWP area during 10 to 25 June 2010.
Light grey is period simulated with 2012 boundary conditions; Dark grey is period simulated with 2010 boundary conditions. . . 77 Figure 31 – Vertical profile of the Virtual Temperature. Dotted lines are day 17 June, light
grey is 2012 boundary conditions; dark grey is 2010 boundary conditions; solid line is the averaged period of 10 to 25 June 2010; light grey is 2012 boundary conditions; dark grey is 2010 boundary conditions. . . 77 Figure 32 – Vertical profile of the Vertical motion. Dotted lines are day 17 June, light
grey is 2012 boundary conditions; dark grey is 2010 boundary conditions; solid line is the averaged period of 10 to 25 June 2010; light grey is 2012 boundary conditions; dark grey is 2010 boundary conditions. . . 78 Figure 33 – Time series of Latent Heat averaged in SAWP area during 10 to 25 June 2010.
Light grey is period simulated with 2012 boundary conditions; Dark grey is period simulated with 2010 boundary conditions. . . 79 Figure 34 – Vertical profile of the Water Vapour. Dotted lines are day 17 June, light grey
is 2012 boundary conditions; dark grey is 2010 boundary conditions; solid line is the averaged period of 10 to 25 June 2010; light grey is 2012 boundary conditions; dark grey is 2010 boundary conditions. . . 80
2010. Light grey is period simulated with 2012 boundary conditions; Dark grey is period simulated with 2010 boundary conditions. . . 81 Figure 36 – Vertical profile of the Wind Speed. Dotted lines are day 17 June, light grey is
2012 boundary conditions; dark grey is 2010 boundary conditions; solid line is the averaged period of 10 to 25 June 2010; light grey is 2012 boundary conditions; dark grey is 2010 boundary conditions. . . 81 Figure 37 – Daily rainfall Hovmöller (longitude 35◦S and averaged latitude between 5◦S
to 10◦S) for period simulated by: a) 2010 and b) 2012 SST boundary conditions. 82 Figure 38 – Mean condition for SST in a) High SST conditions, b) Low SST conditions,
c) near Normal and d) Entire period averaged. . . 84 Figure 39 – Atmospheric integrated water vapour in a) High SST conditions, b) Low SST
conditions, c) near Normal and d) Entire period averaged. . . 85 Figure 40 – Water vapour profile in SAWP area. Dotted light grey mean condition; Dotted
dark grey near normal condition; Solid ligh grey is low SST conditions; Solid dark grey is the high SST conditions. . . 86 Figure 41 – Virtual temperature profile in SAWP area. Dotted light grey mean condition;
Dotted dark grey near normal condition; Solid ligh grey is low SST conditions; Solid dark grey is the high SST conditions. . . 87 Figure 42 – Longitudinal profile of atmospheric water vapor (upper panel), 10-days period
average removed, and ocean temperature (bottom panel) from 20 (a), 21 (b), 22 (c) February 2009 and 30 (d) June, 01 (e), 02 (f) July 2009 at 00UTC. Isolines of 0.012kg / kg (upper panel, black line) and oceanic temperature (bottom panel, black line) with emphasis on the isotherm of 20◦C and 28.5◦C. 88 Figure 43 – Longitudinal profile of atmospheric water vapor (upper panel), 10-days period
average removed, and ocean temperature (bottom panel) from 14 (a), 15 (b), 16 (c) April 2011 and 30 (d) April, 01 (e) May, 02 (f) May 2011 at 00UTC. Isolines of 0.012kg / kg (upper panel, black line) and oceanic temperature (bottom panel, black line) with emphasis on the isotherm of 20◦C and 28.5◦C. 89 Figure 44 – Longitudinal profile of atmospheric water vapor (upper panel), 10-days period
average removed, and ocean temperature (bottom panel) for the maximum convection over ENEB region for 2012 (a), 2013 (b), 2014 (c) March 2015 (d), June 2015 (e), 2016 (f). Isolines of 0.012kg / kg (upper panel, black line) and oceanic temperature (bottom panel, black line) with emphasis on the isotherm of 20◦C and 28.5◦C. . . 90
Table 1 – Validations for: 2m air temperature and 2m Relative Humidity for the model experiments in Ocean area . . . 55 Table 2 – Validations for: 2m air temperature and 2m Relative Humidity for the model
experiments in Land area . . . 55 Table 3 – Validations for: 2m air temperature and 2m Relative Humidity for the model
experiments in Coastal area . . . 55 Table 4 – Validations for: u10m, v10m and Wspdfor the model experiments Ocean area . 55 Table 5 – Validations for: u10m, v10m and Wspdfor the model experiments Land area . . 56 Table 6 – Validations for: u10m, v10m and Wspdfor the model experiments Coastal area 56 Table 7 – Difference between simulated and observed values in different rain gauges in
AD Atlantic Dipole
AESA Agência Executiva de Gestão das Águas da Paraíba AGC Atmospheric Global Circulation
APAC Agência Pernambucana de Águas e Clima ARW Advanced Research in WRF
AVHRR Advanced Very High-Resolution Radiometer CNEB Central Northeast Brazil
COARE Coupled Ocean-Atmosphere Response Experiment COAWST Coupled Ocean-Atmosphere-Wave-Sediment Transport CPTEC Centro de Previsão do Tempo e Estudos Climáticos cSEC Central South Equatorial Current
CSET Cloud Systems Evolution in the Trades DCP Data Collect Platform
EMPARN Empresa de Pesquisa Agropecuária do Rio Grande do Norte ENEB Eastern Northeast Brazil
EUMETSAT European Organisation for the Exploitation of Meteorological Satellites EWD Easterly Wave Disturbance
FNL (Final) Operational Global Analysis Data
GOES Geostationary Operational Environmental Satellite INEMA Instituto do Meio Ambiente e Recursos Hídricos INMET Instituto Nacional de Meteorologia
INPE Instituto Nacional de Pesquisas Espaciais ITCZ Inter Tropical Convergence Zone
MCT Model Coupling Toolkit
MMAB Marine Modeling and Analysis Branch MPI Message Passing Interface
MYNN Mellor-Yamada Nino Nakanishi NBC North Brazilian Current
NBCR North Brazil Current Retroflection NBUC North Brazilian Undercurrent
NCEI National Center for Environmental Information NCEP National Centers for Environmental Prediction NEB Northeast Brazil
NNEB Northern Northeast Brazil
NOAA National Oceanic and Atmospheric Administration nSEC Northern South Equatorial Current
OML Ocean Moisture Layer PBL Planetary Boundary Layer
PIRATA Prediction and Research Moored Array in the Tropical Atlantic ROMS Regional Oceanic Model System
RRTMG Rapid Radiative Transfer Model
RTG-SST Real-time, global, sea surface temperature SACZ Southern Atlantic Convergence Zone SAWP Southern Atlantic Warm Pool
SEC Southern Equatorial Current SF Surface Physics
sSEC Southern South Equatorial Current SNEB Southern Northeast Brazil
TSA Tropical South Atlantic TWIL Trade Winds Inversion Layer
UTCV Upper Troposphere Cyclonic Vortex WPS WRF preprocessing system
WRF Weather Research and Forecasting WTSA Westernmost Tropical South Atlantic
di Model Error Xis Simulated variable
¯
Xis Simulated variable average Xos Measured Variable
¯
Xos Measured variable average
M E Mean error
M AE Mean absolute error M SE Mean squared error RM SE Root mean squared error IA Index of agreement ρa Surface air density cpa Dry air specific heat
Ch Transfer heat exchange coefficient
S Mean wind Speed
Ts Surface temperature θ Potential temperature Lc Evaporation latent heat
Ce Transfer moisture exchange coefficient qs Near surface moisture rate
q Water vapour
Cd Transfer wind stress exchange coefficient us Surface ocean stream flow
u u-wind component
RMsf c Near surface moisture rate Psf c Surface pressure
F LQC Moisture coefficient of exchange
ρ Air density
Mavail Available humidity U∗ u∗in similarity theory κ Von Karman constant P SIQ Humidity resistance
Qf x Upward motion of water vapour
qv First η level above surface Water Vapour LHV Water Evaporation Latent Heat
θgb Near ground potential temperature F LHC Heat coefficient of exchange HF X Upward heat flux
1 INTRODUCTION AND OBJECTIVES . . . . 21
2 THE STUDY AREA CHARACTERISATION . . . . 24
2.1 GEOGRAPHIC DOMAINS . . . 24
2.1.1 Northeast Brazil . . . . 24
2.1.2 Tropical South Atlantic ocean . . . . 25
2.2 TRADE WINDS INVERSION LAYER . . . 27
2.3 EASTERLIES WAVE DISTURBANCES . . . 28
2.4 OCEAN ATMOSPHEREINTERACTION . . . 31
2.5 OCEAN ATMOSPHEREMODELLING. . . 33
3 DATA AND METHODS . . . . 39
3.1 DOMAIN CONFIGURATION ANDDATA USED . . . 39
3.2 PERIOD OF STUDY . . . 40
3.3 EXPERIMENTS . . . 41
3.3.1 Validations and 2010 Extreme Event . . . . 41
3.3.2 Test Case . . . . 41
3.4 STATISTICS . . . 42
3.5 PARAMETERS SEQUENCE . . . 43
4 RESULTS . . . . 48
4.1 VALIDATION . . . 48
4.1.1 Air Temperature and Relative Humidity 2m . . . 48
4.1.2 Wind components at 10m . . . . 50
4.1.3 Statistics Results . . . . 54
4.1.4 Precipitation . . . . 56
4.2 EXTREME RAINFALL EVENT IN 2010 . . . 57
4.2.1 SAWP . . . . 57
4.2.2 Atmosphere–ocean interactions . . . . 60
4.2.2.1 Experiments without sensible and latent fluxes . . . 60
4.2.2.2 Experiments with sensible and latent fluxes . . . 60
4.2.2.3 Coupled modeling experiments with different boundary conditions . . 61
4.2.3 Synoptic analysis of the coupled modeling . . . . 61
4.2.4 Zonal water vapour fluxes . . . 66
4.2.5 Atmospheric vertical structure . . . 66
4.2.6.2 Period 16 to 19 June 2010 . . . 72 4.2.6.3 Period 19 to 25 June 2010 . . . 73 4.3 DIFFERENT SST INPUT . . . 73 4.3.1 Synoptic Comparison . . . . 74 4.3.2 MABL conditions . . . . 76 4.3.3 Rainfall Conditions . . . . 81 4.4 MEAN CONDITIONS . . . 83 4.4.1 Averaged Pattern . . . . 83 4.4.2 Synoptic Patterns . . . . 86
5 CONCLUSION AND PERSPECTIVES . . . . 92
1 INTRODUCTION AND OBJECTIVES
The Northeastern Brazilian region (NEB) covers the area between 47◦W-35◦W; 18◦ S-1◦S (RAO; LIMA; FRANCHITO, 1993) and is divided into different climatic regions. The NEB presents pronounced space-time rainfall variability as a result of different atmospheric forcing (RAO; LIMA; FRANCHITO, 1993; HASTENRATH, 2012; HÄNSEL et al., 2016). This inter annual variability in rainfall is mainly controlled by the sea-surface temperature (SST) of the tropical Pacific, associated with El Niño/La Niña events, (ARAGÃO, 1998; ANDREOLI; KAYANO, 2006; ANDREOLI; KAYANO, 2007) and Atlantic (MOURA; SHUKLA, 1981; MOURA et al., 2009; SILVA; GUEDES, 2012), as well the action of both oceans (ANDREOLI; KAYANO, 2006; ANDREOLI; KAYANO, 2007) causing extreme drought/rainfall in this region. The rainfall variability in the Northern and central semiarid regions of the NEB is associated with the north-south migration of the Inter Tropical Convergence Zone (ITCZ) (MOSCATI; GAN, 2007; HASTENRATH, 2012; RODRIGUES et al., 2011; MARENGO et al., 2017). In the southern, western and central regions of the NEB, the South Atlantic Convergence Zone (SACZ) brings rainfall to the semiarid region during austral spring and summer (NOGUES-PAEGLE; MO, 2002). During January and February there is the occurrence of Upper Troposphere Cyclonic Vortex (UTCVs). In the eastern NEB (ENEB) the precipitation is modulated by Easterly Waves Disturbances (RAMOS, 1975; TORRES; FERREIRA, 2011; KOUADIO et al., 2012; GOMES et al., 2015) with maximum rainfall between May and July, with annual average rainfall above than 1500 mm.
In 2010, a succession of intense rainfall events given rise to floods in the eastern rivers of the Pernambuco and Alagoas States. The World Bank report and the government of Pernambuco (2010), registered a total of 67 damaged cities, 20 deaths, almost 30,000 homeless and an economic loss of approximate $ 1 billion. In that year, it was possible to observe positive SST anomalies in the Westernmost Tropical South Atlantic (WTSA) which exceeded 1◦C in relation to the climatology between February and June, recorded in the Prediction and Research Moored Array in the Tropical Atlantic (PIRATA) (SERVAIN et al., 1998) buoy located at 8◦S 30◦W, as well as the SST data from the Advanced Very High-Resolution Radiometer AVHRR (http://oceanwatch.pifsc.noaa.gov).
There is a positive correlations between SST over the westernmost TSA regions and the variability of NEB precipitations (MOURA et al., 2009). These correlations are connected to water vapour transport from ocean to land (CAVALCANTI; GANDU; AZEVEDO, 2002), heat and moisture exchange between ocean and atmosphere (FOLTZ; MCPHADEN, 2006; CINTRA et al., 2015; HOUNSOU-GBO et al., 2015) and instabilities over warm waters (WANG; ENFIELD, 2001). Climatically the WTSA thermal seasonality has a small variation (±2◦C) as well as low average monthly anomalies (±0.3◦C) (HOUNSOU-GBO et al., 2015). Due to SST
anomalies exceed 28.5◦C in period from February to June in WTSA, these regions were similar to a Warm Pool or Southern Atlantic Warm Pool (SAWP) (WANG; ENFIELD, 2001; WANG et al., 2006; WANG; LEE, 2007).
Oceanic Warm Pools are regions with temperatures above 28.5 ◦C with well defined seasonal patterns in terms of time and space (WANG; ENFIELD, 2001). In these regions, oceanic dynamics favours low pressure on the surface, moisture convergence, increasing the water vapour content, atmospheric instabilities and convective cloudiness. Besides, intensifying the meteorological systems that will modulate the rainfall in continental regions adjacent to the warmer surface waters (BROWN; ZHANG, 1997; WANG et al., 2006; KOUADIO et al., 2012; HOUNSOU-GBO et al., 2015).
Over the entire trade winds regions around the Earth, there is a stable layer in the low troposphere resulted from the equilibrium between subsidence and the surface fluxes, which isolates the well mixed Marine Atmospheric Boundary Layer (MABL) from the free atmosphere above and inhibits the vertical motion (LINDZEN; NIGAM, 1987). This layer is called Trade Winds Inversion Layer (TWIL), and it is characterised by a strong increase of the stable vertical gradient of potential temperature and a high decrease of the mixing ratio (JOHNSON et al., 1999). This dynamics are presented either in Tropical South Atlantic and NEB region.
The trade winds inversion height is controlled horizontally by averaged values of sea surface temperature, atmospheric divergence and above-inversion atmospheric structure (SCHU-BERT et al., 1995). Due to this, in regions over oceanic warm pools it is common found higher TWIL base, where can be found in 925 to 800 hPa (SCHUBERT et al., 1995; CARRILLO et al., 2015), and these effect cause many of convective rainfall over regions nearby warm waters (JOHNSON et al., 1999). Hence, convective precipitation over the ENEB has a direct correlation with the height of the base of the thermal inversion layer, which, in turn, is correlated with the SST on WTSA.
The interaction between air and sea, in SAWP region, is important for understanding the climatic variability over ENEB, due to heat, momentum and moisture exchange between these two media (BOURRAS et al., 2004). Ocean surface fluxes are determined by ocean surface temperature as well near-surface air temperature, humidity and wind speed, all of which are explicitly linked to MABL processes (ZENG et al., 2004). As the MABL interact with more warmer (colder) waters, both latent and sensible heat fluxes increases (decreases), the TWIL weakens (strengthens) and there are higher (smaller) well mixed MABL (PYATT et al., 2005).
The vertical structure of MABL is a key factor to triggering the occurrence of some synoptic activities such as deep convection (PENG et al., 2016). The warmer SST over SAWP region is directly related to rainfall variability over Recife (ENEB) (HOUNSOU-GBO et al., 2015). But, such as the warmer (colder) and large (small) warm pool is linked to more (less) intensity and formation of hurricanes in the North Atlantic (WANG et al., 2006), the SAWP region, also, can be an important region to intensify meteorological systems over ENEB leading
to extreme rainfall.
The Easterly Waves Disturbances, which modulates the ENEB rainfall are observed during entire year, but have more capability to generate rainfalls in the austral winter, giving rise to high daily precipitations (KOUADIO et al., 2012; GOMES et al., 2015). The ascending humidity from low to medium levels of the atmosphere occurs due to the presence of upward movements, moisture convergence, thermal fluctuation, and cyclonic vorticity in the 850 hPa, 700 hPa and even 500 hPa levels (KOUADIO et al., 2012; GOMES et al., 2015). Due to these synoptic-scale meteorological disturbances there is a weakening, or breaking, of the TWIL which allows the well mixed MABL provide moisture to mid and higher levels of the atmosphere (GRAY, 1968; LINDZEN; NIGAM, 1987; KORACIN; ROGERS, 1990; SCHUBERT et al., 1995; JOHNSON et al., 1999; PYATT et al., 2005).
Due to a lack of data set over TSA, it is difficult to study the air-sea interactions over the SAWP region during meteorological disturbances, which cause extreme rainfall episodes in the ENEB. To fill this gap, simulations with regional coupled atmosphere and ocean models can be performed. Three-dimensional coupled atmosphere-ocean models have been developed and applied to idealised and realistic scenarios, in order to predict the interactions between a tropical cyclone and the ocean (WARNER et al., 2010). Also, the expectations in coupling atmosphere and oceanic models are the well resolved interactions between the physics in either MABL and Ocean Mixed Layer (OML) to generate more precisely intense weather conditions to the forecast (RENAULT et al., 2012).
The Coupled Ocean-Atmosphere-Wave-Sediment Transport (COAWST) is comprised of several components that include models for the ocean, atmosphere, surface waves, sediment transport, a coupler to exchange the data fields, and a method to regridding (WARNER et al., 2010). The system has interactivity with all components, exchanging prognostic variables between them, allowing more realistic resolving in air-sea feedback interaction than only an atmospheric model which uses static SST (WARNER et al., 2010; RENAULT et al., 2012).
The main objective of this work is to investigate the sea-air interaction processes over the SAWP and the relationship with extreme rainfall events. Also, there is a interest in verify the capability of the coupled model to predict intense rainfall over the ENEB. Identify the effects of ideal different SST conditions in the model convection. Quantify the impacts in atmospheric thermodynamics under different SAWP conditions. Verify the influence of SAWP in different real conditions.
2 THE STUDY AREA CHARACTERISATION
2.1 GEOGRAPHIC DOMAINS 2.1.1 Northeast Brazil
The Northeast Brazil is sited in tropical region between 1◦S - 18◦S Latitude ; 35◦W - 47 ◦
W Longitude. In all its extension, the northeast covers localities with rainfall above 1500 mm, in coastal areas, and regions with rainfall climatology about 400 mm annually (KOUSKY, 1979; KOUSKY, 1980; RAO; LIMA; FRANCHITO, 1993). Due to climatic and physical discrepancy, the NEB has humid and semi-arid climates (also, Sertão), with transition regions known as Agreste. Due to topographic features (Figure 1), there are local circulation factors (FEREIRA; MELO, 2005) that may generate micro climates in higher topography regions, such as the Chapada Diamantinaand Chapada do Araripe, which are humid climates in semiarid border regions. Figure 1 shows the Brazilian topography.
The high variability of NEB is characterised by different atmospheric synoptic activity, leading three (in average) areas with different rainfall climatology (KOUSKY, 1979) as shown in Figure 2. In the northern and central semiarid regions of the NEB the north-south migrations of the Inter Tropical Convergence Zone (ITCZ) reach the maximum precipitation between February and May (MOSCATI; GAN, 2007; HASTENRATH, 2012; RODRIGUES et al., 2011; MARENGO et al., 2017). The hydric excess in the northern NEB (NNEB) is correlated with the ENSO cold phase (ANDREOLI; KAYANO, 2006; ANDREOLI; KAYANO, 2007) and with negative Atlantic Dipole (AD), which means positive Tropical South Atlantic (TSA) SST anomalies (MOURA; SHUKLA, 1981; KOUADIO et al., 2012; HOUNSOU-GBO et al., 2015). In the southern (SNEB), western (WNEB) and central (CNEB) regions of the NEB, the cold fronts (KOUSKY, 1979) and the South Atlantic Convergence Zone (SACZ) brings rainfall to the south of semiarid region during austral spring and summer (NOGUES-PAEGLE; MO, 2002). The SACZ is characterised by an organised Convective Cloud-Band (CCB) that generally extends from the Amazon to the Atlantic Ocean in a northwest-southeast axis (ROBERTSON; MECHOSO, 2000).
In the eastern NEB (ENEB) the precipitation is modulated by Easterly Waves Distur-bances (EWD) (RAMOS, 1975; TORRES; FERREIRA, 2011; KOUADIO et al., 2012; GOMES et al., 2015) with maximum rainfall between May and July, with annual average rainfall above than 1500 mm. The TSA regions near the east coast of the NEB are responsible for the supply of water vapour, which are carried by the southeast trade winds (CAVALCANTI; GANDU; AZEVEDO, 2002) into the continent. Thus, the ENEB has a positive correlation between the rain-fall anomalies and the Western South Tropical Atlantic (STA) SST anomalies (HOUNSOU-GBO et al., 2015).
Figure 1 – Northestern Brazil topography. Near coastal areas elevation in 0 - 200m height, meanwhile central areas varying from 200 - 1200m in height.
Source: Instituto Brasileiro de Geografia e Estatística - IBGE
During December to February there is the occurrence of Upper Troposphere Cyclonic Vortex (UTCVs). These vortex are formed and/or intensified downstream from strongly amplify-ing mid-latitude frontal systems, and are maintained by a direct thermal circulation, with cold air sinking in the centre and relatively warm air rising on the periphery (FEREIRA; MELO, 2005). Figure 2 shows the zonal climatology of Brazilian Northeast.
2.1.2 Tropical South Atlantic ocean
In similar manner, the dynamics that control the variability in the western boundary regime of the TSA could be a result of the seasonal changes of the surface wind speed. The ITCZ migrates meridionally, on seasonal time scale, leading to changes in the upper ocean parameter distribution and also in the ocean circulation (STRAMMA et al., 2005). A major portion of variability in the tropical Atlantic currents seems to be driven by large-scale seasonality in the trade wind regime and the latitude of the ITCZ (STRAMMA; ENGLAND, 1999; LUMPKIN;
Figure 2 – Brazilian Northeast rainfall climatology zones. I - Is the Southern Northeast Brazil, with maximum precipitations between November and February; II - Northern North-east Brazil, with maximum precipitation between February and May; III - Eastern Northeast Brazil, maximum precipitation between May and August.
Source: Adapted from IBGE
GARZOLI, 2005).
The Southern Equatorial Current (SEC) forms the northern part of the South Atlantic Ocean subtropical gyre, carrying subtropical waters from the Benguela Current region towards the Brazil shelf region around 14 ◦S, where it bifurcates into the North Brazil Undercurrent (NBUC), later North Brazilian Current (NBC) to the north and the Brazil Current (BC) to the South (RODRIGUES; ROTHSTEIN; WIMBUSH, 2007). The SEC can be divided into three bands in the South Atlantic. The northern SEC (nSEC), the central SEC (cSEC) and the southern SEC (sSEC) flows (STRAMMA; ENGLAND, 1999). Figure 3 represents the schematic representation of mean current flow in tropical Atlantic.
In particular, the sSEC bifurcation region brings together multiple oceanic weather and climate interactions of great importance. Among these interactions they can be enumerated: (i) transfers of heat and mass between different layers of the tropical Atlantic subsurface; (ii) exchanges of heat and fresh water between the ocean and the atmosphere on the tropical Atlantic surface; (iii) links between climatic variability of the SST and the heat content of the upper layers of the tropical Atlantic and related atmospheric systems, which controls precipitation on
Figure 3 – Schematic representation of mean zonal and meridional currents are indicated: The upper ocean northward NBUC flow (red solid line), and the deepest southward transport by the DWBC (blue dashed line). The sSEC bifurcation at 100m depth (solid line); between 200m and 500m depth (dashed line); between 500m and 1200m depth (pointed line), according with (STRAMMA; ENGLAND, 1999).
Source: Veleda, 2008
the NEB.
These circulation characteristics lead to a thermal climatology over TSA. The ocean water are colder in eastern side of basin due to upwelling and a warmer water in western side due to downwelling. The seasonal cycle of temperature shows values varying from about 25.5◦to 28.0◦ from September to April in longitude near 34◦W. After reach the maximum temperature, there is a decreasing, reaching the minimum in Semptember with temperatures near then 25.5 ◦. In longitude near 20◦W, the same patterns are presented but temperatures are colder with thermal amplitude of 23.5◦C to 26.5◦C. The pattern of cooling of western basin of TSA are due to latent heat losses (FOLTZ; MCPHADEN, 2006), and coincide with the most rainfall period in ENEB (CINTRA et al., 2015) which are, also, direct related with rainfall anomalies in this areas (MOURA et al., 2009; KOUADIO et al., 2012; HOUNSOU-GBO et al., 2015).
2.2 TRADE WINDS INVERSION LAYER
The Atmosphere Global Circulation (AGC) is caused by solar irradiance which cause a meridional gradient of temperature, where are warm areas near the equator and cold areas in the poles.
Due to the equator receive more heat then high latitudes regions, there is a upward motion of the air and a surface low pressure. At the Equatorial region in the top of troposphere, the air diverges and generates a poleward wind flow descending at the middle latitudes.
The subsidence at mid Latitude generates a surface high pressure, causing a pressure gradient between mid latitudes and equator generating wind flows equatorward.
Then, the Coriolis effect turns the wind due to earth circulation, causing persistent wind at surface with southeast (northeast) direction in south (north) hemisphere called Trade Winds.
The thermodynamics of trade winds shows a inversion layer, in trade wind belt, caused by descendent branch of Hadley cell and local thermodynamics patterns (SCHUBERT et al., 1995; CARRILLO et al., 2015) here called Trade Winds Inversion Layer (TWIL) as shown in Figure 4. The base of TWIL is characterised by a strong increase of the stable vertical gradient of potential temperature and a high decrease of the mixing ratio (JOHNSON et al., 1999). These inversions acts as a barrier layer preventing upward motion of humidity and high vertical cloudiness from a moist and buoyant Marine Atmospheric Boundary Layer (MABL) (LINDZEN; NIGAM, 1987).
Therefore TWIL regions are connected to clear or cumulus humilis clouds types over warm waters and to high stratiform cloud cover over cold waters (CARRILLO et al., 2015) as exemplified Figure 4. It’s verified warm rains in regions influenced by trade winds surrounded by warm waters due to development of tower cumulus (LIU; ZIPSER, 2009), similar to ENEB commonly daily rainfall. In fact, where the warm water are presents and a deep TWIL is observed, cumulus towers can appears with the cloud top height limited to the TWIL base.
Over the oceans where the SST gradients are present, the TWIL shows an important characteristic. Since over cold waters the shallow and mixed MABL are under influence of a high inversion, there are high stratiform cloud cover. However, crossing the SST gradient, these structure can be broken, by entrainment and turbulence, its make the regions uncoupling stratiform clouds and change to cumuliform. These shallow culuniform clouds when crossing the warm ocean acquires energy from ocean being more high developed (Tower Cumulus) limited by base of TWIL (PYATT et al., 2005; CARRILLO et al., 2015).
The example of how it occurs are show in the Figure 5. Near Africa west coast are present stratiform clouds in cold waters region. In Central South Atlantic there is a decoupling region, with stratiforms and cumuliform clouds. Near the East coast of Brazil, the tower cumulus near land.
2.3 EASTERLIES WAVE DISTURBANCES
Easterlies Wave Disturbances (EWD) are atmospheric disturbances which travels in trade winds easterlies and in the easterlies of equatorial through (also known as ITCZ). EWD are responsible to brings huge quantity of moisture, tropical instability and upward motions causing development of cumulus cloudiness and responsible to module the rainfall regime over
Figure 4 – Trade Wind inversion layer Structure
Source: Cloud Systems Evolution in the Trades - CSET
Figure 5 – Satellite image. Day 08 January 2018 at 0000UTC.
the influenced regions. (FRANK, 1969; REED; NORQUIST; RECKER, 1977; DIEDHIOU et al., 1998; KAYANO, 2003; TORRES; FERREIRA, 2011; GOMES et al., 2015).
In the dynamical structure and synoptic-scale pattern, the EWD presents convergence (divergence) in low levels in vicinity of through (ridge), high (low) relative humidity and upward (downward) motion over the trough (ridge) (CARLSON, 1969; REED; RECKER, 1971; REED; NORQUIST; RECKER, 1977; TAI; OGURA, 1987; DIEDHIOU et al., 1998; KAYANO, 2003; DIEDHIOU; MACHADO; LAURENT, 2010; TORRES; FERREIRA, 2011; GOMES et al., 2015). Over the through, linked to upward motion, there is a pattern of convective cloudiness in shape of inverted "V" in northern hemisphere, opposite in south hemisphere, and cloud blobs (FRANK, 1969).
These characteristics persists over the entire tropical regions and make possible to identify these disturbances in many ways, such as: Satellite composite (CARLSON, 1969; FRANK, 1969; YAMAZAKI; RAO, 1977), Wind and satellite composite (REED; RECKER, 1971; REED; NORQUIST; RECKER, 1977; REED; KLINKER; HOLLINGSWORTH, 1988; DIEDHIOU et al., 1998; GOMES et al., 2015), Spectral analysis (DIEDHIOU et al., 1998; KAYANO, 2003; DIEDHIOU; MACHADO; LAURENT, 2010), Dynamical diagrams (Hovmuller) (NEVES; ALCÂNTARA; SOUZA, 2016) and others techniques (THORNCROFT; HODGES, 2001; BERRY; THORNCROFT; HEWSON, 2007; SERRA; KILADIS; HODGES, 2010; TORRES; FERREIRA, 2011).
There is EWD over entire tropical areas in the entire world but they vary seasonally in strength and intensity (CARLSON, 1969) and have more records in northern than southern hemisphere (YAMAZAKI; RAO, 1977; TAI; OGURA, 1987). Some of the EWD are responsible to develop the tropical cyclonic around the world, such as Hurricane in the North America and Typhoon in the the West Pacific basin. The African Easterlies Waves (AEW) is an example of these tropical disturbances (CARLSON, 1969; FRANK, 1969; BURPEE, 1972; REED; NORQUIST; RECKER, 1977).
The AEW are disturbances developed in Saharan continent and crosses the Tropical North Atlantic ocean (FRANK, 1969; BURPEE, 1972), and many of those AEW develop Central and North America tropical storms and hurricanes (CARLSON, 1969). In western Pacific the EWD are less intense than AEW, but the occurrence and the displacement occurs in the same way, crossing tropical Pacific reaching the Eastern coast of continents in the West Pacific region, sometimes developing to a Typhoon (TAI; OGURA, 1987).
The influences of EWD in Tropical South Atlantic (TSA) was observed by (YAMAZAKI; RAO, 1977) with satellite composite and noticed seasonal types of clouds moving eastward by trade winds, but with less intensity than another places in earth. Also using Longitude x time diagrams from meridional wind component, is possible to perceive the presence of the EWD over the ENEB (NEVES; ALCÂNTARA; SOUZA, 2016). These disturbance can occur in every seasons, but to appear more organized during March-April-May (MAM) and
June-July-August (JJA) where there is more variations in meridional wind, there is more activity of EWD in this period (COUTINHO; FISCH, 2007; DIEDHIOU; MACHADO; LAURENT, 2010), coinciding with the wet season over the ENEB (KOUSKY, 1979; KOUSKY, 1980; RAO; LIMA; FRANCHITO, 1993).
Each EWD over STA, as well around the world, have different dynamical patterns compared to each other. For example, the EWD moving around equator have more phase speed than southern ones, the period can vary season to season where the double of records can be observed between dry and wet season (COUTINHO; FISCH, 2007), also the high and deepens of through where the trough can be found on 925, 850, 700 or 500 hPa and wavelength (KAYANO, 2003; COUTINHO; FISCH, 2007; GOMES et al., 2015), and this cause variations in the intensity of rainfall under EWD influence.
2.4 OCEAN ATMOSPHERE INTERACTION
The ocean and the atmosphere has a strictly relationship to control the climate on Earth. Heated by the sun’s radiation, the ocean and land surface evaporate water, which then moves around with wind in the atmosphere, condenses to form clouds, and falls back to Earth’s surface as rain or snow, with the flow to oceans via rivers completing the hydrology cycle (TRENBERTH, 2011).
The ocean is an important component of the climate system. It provides the surface temperature boundary condition for the atmosphere over 70% of the globe. It absorbs over 97% of solar radiation incident on it from zenith angles less than 50◦. It provides 85% of the water vapour in the atmosphere. It exchanges, absorbs and emits a host of radiative important gases. It is a major natural source of atmospheric aerosols. Thus, even a static ocean would significantly influence the climate. However, the ocean is dynamic and its surface properties will vary on all time scales, allowing great scope for feedback between the ocean and atmosphere (BIGG et al., 2003).
The Marine Atmospheric Boundary Layer (MABL) over the Earth’s ocean plays a critical role in regulating the surface energy and moisture fluxes and in controlling the convective transfer of energy and moisture to the free atmosphere (KLOESEL; ALBRECHT, 1989; PYATT et al., 2005). Ocean surface fluxes are determined by ocean surface temperature as well as near-surface air temperature, humidity and wind speed, which is explicit linked to the MABL (ZENG et al., 2004). So, the structure of MABL, which contains most of water vapor and energy flux in atmosphere column, is one of the key factor that trigger the occurrence of some synoptic activities, such deep convection (KLOESEL; ALBRECHT, 1989; PENG et al., 2016).
Across the ocean, the tropical western basin are warmer than eastern side due the mid-latitude gyre and air-sea interactions, mainly subtropical high (CLEMENT; SEAGER; MURTUGUDDE, 2005). Its cause a downwelling (upwelling) of waters lead to regions generally
covered by waters warmer (colder) in west (east) ocean basins. Sometimes, where ocean basins reaches temperatures higher than 28.5◦C, its called Warm Pools (WANG; ENFIELD, 2001). The process of the formation and maintenance of warm pools is the ocean-atmosphere feedback, such as convective activity and cloudiness resulting less net long-wave radiation loss (CLEMENT; SEAGER; MURTUGUDDE, 2005; WANG; ENFIELD, 2003).
Warmer SST are associated with warmer and moister troposphere, reducing Sea Level Pressure (SLP), weaker easterlies winds, less vertical wind shear, and weakened subsidence aloft. Years with well developed warm pool, these conditions lead to generally greater summer rainfall over Intra American Sea (IAS), Central America (WANG; ENFIELD, 2003) and rainfall regimes over Western Pacific Regions (BROWN; ZHANG, 1997), and the opposite also occurs (WANG; ENFIELD, 2003; BROWN; ZHANG, 1997), the cold water generate dry summer in these regions.
There is two key factors to lead the storm genesis and intensification. Due to favourable effects of latent and sensible heat fluxes, derived from the ocean surface layers, and the vertical wind shear in the troposphere (DAILEY et al., 2009). These factor are present in warmer warm pool conditions. For example, the North Atlantic Warm Pool largely reduces the troposphere vertical wind shear and increases latent heat losses during August to October (WANG; LEE, 2007). Also the tropical cyclone intensity increases when more heat content are available in deeper ML (CHAN; DUAN; SHAY, 2001).
Also, the track and intensity of hurricanes are modulated by chain of warm SST. The influence of landfall precipitation between tropical disturbances and storms in different places in United States of America (USA), shows the amount and route of storms are variants in cool and warm SST North Atlantic. Even so, the hurricanes developed over Gulf of Mexico in cool SST are weaker then eastern Atlantic warm waters due to short pathways of warm waters (DAILEY et al., 2009).
However, the structure of MABL and the TWIL height also perform important issues to climate modulation over tropical regions. For example, the displacement speed of TC is responsible to modify his structure. In the presence of a strong advective flow, the TC has less time to be cooled by the ocean, and thus the TC becomes more intense (CHAN; DUAN; SHAY, 2001). Also, the height of TWIL is a limiting factor to the development of high cumulus clouds (SCHUBERT et al., 1995). These convective clouds patterns limited by TWIL are associated with develops of warm rains (LIU; ZIPSER, 2009).
Analogously to TC dynamics, the atmospheric disturbances over south Atlantic have more convective activity over the ENEB in the presence of warmer waters over Westernmost Tropical South Atlantic (WTSA) (MOURA et al., 2009; KOUADIO et al., 2012; HOUNSOU-GBO et al., 2015). In the rainy period of ENEB, when EWDs interact with local circulation, low-level convergence can be increased, leading to high amount of precipitation (GOMES et al., 2015). These low-level convergence is increased by warmer waters surface in WTSA (KOUADIO
et al., 2012).
The EWDs has high frequency in the layers in the 850-700 hPa layers during the rainy season (COUTINHO; FISCH, 2007). The propagation of EWDs to the ENEB is associated with center of moist and upward motion associated to a humidity convergence in low levels (GOMES et al., 2015). These 850-700 hPa layers is common to find the base of TWIL (SCHUBERT et al., 1995). So these disturbances in 850-700 hPa acts to provide low level moisture to mid-levels, above the inversion layer, which can favors to the deep convection.
Therefore in periods where anomalies of surface waters in WTSA is higher than 28.5◦ the west basin acts as a warm pool, its commonly named as SAWP (HOUNSOU-GBO et al., 2015). The MABL structure, the air-sea interaction over SAWP and the influence in the rainfall over ENEB is important due to impacts in flash floods in this regions. In scenario with climate change, the characterization of MABL, in the coupled ocean atmosphere forecasting models, can be a tool for mitigating the negative effects of extreme rainfalls.
2.5 OCEAN ATMOSPHERE MODELLING
Due to the climate vulnerability of ENEB, improvements in atmospheric modeling techniques are constantly employed (TORRES; FERREIRA, 2011). Because it has a direct relationship with the TSA, it is also well known that the numerical resolutions of the atmosphere ocean interface is a strategy for better predictions of intense rainfall episodes.
With the growing of computational power, the coupled modeling approach can provides means to develop multidisciplinary application with more robustness. The coupled model is capable to interact in two-way between the models either operating on different scales, or simulating different set of interdependent processes (WARNER; PERLIN; SKYLLINGSTAD, 2008).
The coupled ocean–atmosphere–wave and sediment transport (COAWST) modeling system is comprised of several components that include models for the ocean, atmosphere, surface waves, sediment transport, a coupler to exchange data fields, and a method for remapping, using all publicly available components (WARNER et al., 2010).
The model coupling toolkit (MCT) is an open software library for constructing parallel coupled models from individual parallel models. This application contains multiple instances of the MxN problem, the problem of transferring data between two parallel programs running on disjoint sets of processors. MCT was created as a generalized solution to handle these and other common functions in parallel coupled models (JACOB et al., 2005).
The structure built (Figure 6) consists of a main program that serves to control the flow and processor allocation of the coupled model system, called Master. This Master program was coded for run simultaneously on several different processors. At execution, the Master driver program distributes the total number of processors allocated for the job to the individual models,
Figure 6 – A flow chart of the coupled code forming a single-executable modeling system running in concurrent mode.
Source: (WARNER; PERLIN; SKYLLINGSTAD, 2008)
and calls each model component with an initialize, run where is defined the synchronization point, and finalize structure (WARNER; PERLIN; SKYLLINGSTAD, 2008). Figure 6 summarize the flows of routines in single-executable modeling system. This represent a layout of how the coupling model works.
To allow the models to exchange data fields on different grids we use the spherical coor-dinate remapping interpolation package (SCRIP) to compute interpolation weights (WARNER; PERLIN; SKYLLINGSTAD, 2008). The SCRIP provides a mechanism for computing remapping weights for both first- and second-order remapping. Use of these weights provides remapping that are up to second-order accurate and conservative to machine accuracy (JONES, 1999).
Because the method is completely general, weights can be computed for any type of grid on a sphere, allowing developers of component models in a coupled model context to use whatever grid is appropriate for a given component and not be constrained by compatibility with other component model (JONES, 1999).
The build scripts for the individual systems were modified such that the COAWST is compiled to produce one single executable. The modeling system runs in message passing interface (MPI) and before execution the user sets the number of processors that will be allocated for each model (ocean, atmosphere, and wave) (WARNER et al., 2010).
During initialization each processor from each model initializes with MCT, which is identified the distribution of model components on all the processors. MCT can then determine the grid distributions. During execution the models will advance in time until they reach a user defined synchronization point (WARNER et al., 2010). At that time each model fills its attribute vector to exchange prognostic variables through MCT to other models. The variables that are exchanged are shown in Figure 7.
The atmosphere model provides 10-m surface winds (U 10wind, V 10wind) to the wave and ocean models. The atmosphere also provides to the ocean model the atmospheric pressure (P atm), relative humidity (RH), atmospheric surface temperature (T air), cloud fraction (cloud), precipitation (rain), shortwave (SW rad) and longwave (LW rad) net heat fluxes (WARNER et al., 2010).
The ocean model uses atmosphere model parameters in the coupled ocean atmosphere response experiment (COARE) algorithm (FAIRALL et al., 1996) to compute ocean surface stresses and ocean surface net heat fluxes. But, the fluxes calculated in both parameters can be inconsistent, to avoid this, the heat and moisture fluxes are calculated by atmospheric model and give to ocean model (ZAMBON; HE; WARNER, 2014). The ocean model provides SST to the atmosphere model. The ocean provides surface currents (us, vs), free surface elevation (g), and bathymetry (bath) to the wave model (WARNER et al., 2010).
The surface currents are averaged using a formulation of (KIRBY; CHEN, 1989) that integrates the near-surface velocity over a depth controlled by the wave number. The wave model provides significant wave height (Hwave) and wave length (Lwave) to the atmosphere and the ocean models. The atmosphere model uses these wave values to compute an enhanced sea surface roughness (WARNER et al., 2010).
The wave model also provides to the ocean model wave direction (Dwave), surface and bottom periods (T surf , T bott), percent wave breaking (Qb), wave energy dissipation (W dissip), and bottom orbital velocity (U b). These parameters are used by the ocean model in the COARE algorithm to provide an increased surface roughness and in the sediment transport algorithms for bed load transport and bottom stress mobilization (WARNER et al., 2010).
The wave parameters are also available to compute wave-driven flows. The sediment model can provide spatially-varying bottom roughness as different grain sizes are mobilized and transported. The varying roughness is fed back to the ocean and wave models. The model uses the momentum and buoyancy fluxes computed by Weather Research and Forecasting (WRF) to directly drive the ocean model (ZAMBON; HE; WARNER, 2014). Figure 7 shows the possible configurations of data fields exchanged. Lower right panel (case D) shows all the fields that can be distributed. However, for this works, only WRF and ROMS interaction are activated, so for this work the chart flow are the Figure7b.
Figure 7 – A flow chart of the coupled code forming a single-executable modeling system running in concurrent mode.
Source: (WARNER et al., 2010)
system designed for both research and operational applications. WRF reflects flexible, state-of-art, portable code that is efficient in computing environments ranging from massively-parallel supercomputers to laptops.
The Advanced Research in WRF (WRF-ARW) solver implements fully compressible Eulerian nonhydrostatic equations, with prognostic variables such as the u and v velocity components in Cartesian coordinates; vertical velocity w, potential temperature θ, geopotential φ and surface pressure perturbations; a terrain-following, dry hydrostatic-pressure, with vertical grid stretching permitted vertical coordinate η (ETA), horizontal grid based in Arakawa-C grid staggering; time integration in 2nd- or 3rd-order Runge-Kutta time-split integration scheme with smaller time steps for acoustic and gravity-wave modes; 2nd to 6th-order advection options in horizontal and vertical spatial discretization; turbulent mixing and model filters with sub-grid scale turbulence formulation in both coordinate and physical space, divergence damping, external-mode filtering, vertically implicit acoustic step off-centering and explicit filter option; one-way interactive, two-way interactive and moving nests, multiple levels and integers ratios for nesting options (SKAMAROCK et al., 2008).
Figure 8 – Flowchart of parameterization information exchange in WRF.
Source: Jimmy Dudhia
The physics categories are (1) microphysics, (2) cumulus parameterization, (3) planetary bound-ary layer (PBL), (4) land-surface model, and (5) radiation. Physics packages compute tendencies for the velocity components (un-staggered), potential temperature, and moisture fields (SKA-MAROCK et al., 2008).
The directly influence of the COAWST in WRF solving is in the surfaces physics. However, this directly influence cause the chain effect for the entire atmosphere simulation, by sequence of information exchange as in Figure 8. The surfaces physics will provide the predicted variables to PBL and Radiation physics which in turn will have a feedback process with other parameterizations.
The ROMS solves the free surface primitive equations in a rotational earth-centered environment, based on the classical Boussinesq approximation and the vertical momentum hydrostatic balance (SHCHEPETKIN; MCWILLIAMS, 2005). The model considers the coast lines and coordinates that accompany the terrain, which allows to minimize the number of blind spots in the computational solution. The boundary conditions for the model are therefore appropriate for an irregular oceanic bathimetry and shorelines.
Boundary data include the influences of surface wind stress, heat and mixing fluxes, coastal river influences, bottom trawling, and radiative loss in the open ocean and nudging over the specific circulation in the basin. The model has previous adaptations in different regions of the globe (HAIDVOGEL et al., 2000; SHE; KLINK, 2000; PENVEN et al., 2000; PENVEN et al., 2001; MACCREADY; GEYER, 2001; LUTJEHARMS; PENVEN; ROY, 2003) Vertical advections in ROMS are treated with a third-order scheme that enhances the solution by generating abrupt gradients as a function of a given grid size (SHCHEPETKIN; MCWILLIAMS, 1998).
Unresolved processes of vertical sub-grid scales are parameterized by an adaptation of the planetary boundary layer scheme of the K profile (LARGE; MCWILLIAMS; DONEY, 1994). A full description of the model can be found in Haidvogel et. al. (2000).
3 DATA AND METHODS
3.1 DOMAIN CONFIGURATION AND DATA USED
The domain studied (45◦W - 10◦E; 20◦S - 5◦N), extends from the west coast of Africa to the east coast of the South American continent. It comprised the easternmost area of the Brazilian Northeast (ENEB), selected by the blue line (40.5◦W - 35◦W; 15.5◦S - 4◦S) (Figure 9). The western oceanic area (34◦W - 25◦W; 12.5◦S - 4◦S) characterise the Southern Atlantic Warm Pool (SAWP) region, selected by hatched square; and the 20◦W - 10◦E; 20◦S - 8◦S domain, is the region of thermal gradients of the South Atlantic in larger squared with crosses hatched.
The meteorological fields were analysed by Geostationary Operational Environmen-tal Satellite (GOES) from National Oceanic and Atmospheric Administration (NOAA) and MeteoSAT from European Organisation for the Exploitation of Meteorological Satellites (EU-METSAT) satellite images, as well sounding profiles of the Recife city - PE (35.1◦W; 8.5◦S), from Instituto Nacional de Meteorologia (INMET). Precipitation data were available from Agên-cia Pernambucana de Águas e Clima(APAC), yellow dots; the Agência Executiva Gestão das Águas da Paraíba(AESA), magenta dots; the Empresa de Pesquisa Agrícola do Rio Grande do Norte(EMPARN), blue dots; the Instituto de Meio Ambiente e Recursos Hídricos (INEMA).
Other meteorological observations are from the Data Collection Platforms, or DCPs, monitored by INMET (Figure 9 - red dots). For ocean modeling, we use in situ data from the oceanographic buoys of the PIRATA (Figure 9 - green dots) (SERVAIN et al., 1998) (available at: http://www.pmel.noaa.gov/Pirata/), located in the domain of study.
For the atmospheric modeling, two different databases were used, the boundary data were the NCEP FNL (Final) Operational Global Analysis Data (NCEP, 2000), with a resolution of 1◦ x 1◦ and the ERA-Interim Global Atmospheric Reanalysis (DSS / CISL / NCAR / UCAR / ECMWF 2005) with resolution of 0.7km (N128) (http://www.ecmwf.int), both every 6 hours.
When the coupled model between the ROMS model and the WRF was used, the variables of temperature, salinity and currents for the oceanic model were given by the Hybrid Coordinate Ocean Model (HYCOM), with resolution of 0.08◦ x 0.08◦ (WALLCRAFT et al., 2008).
When the coupling between WRF and ROMS was disconnected, the oceanic boundary condition was provided by the "Real-time, global, sea surface temperature (RTG-SST) analysis", developed by the National Centers for Environmental Prediction / Marine Modeling and Analysis Branch (NCEP / MMAB). RTG-SST has a resolution of 0.5◦x 0.5◦and was interpolated every 6 hours.
Figure 9 – Studied Domain. Squared with lines is the SAWP region. Squared with crosses hatched is the Gradient regions. Greens dots is the PIRATA buoy. Red dots is the INMET PCD. Yellow dots is APAC pluviometers. Magenta dots is AESA pluviometers. Blue dots is EMPARN pluviometers. Shaded color is the Earth elevations (bathymetry and topography).
Source: The author. ENEB Region delimited by (RAO; LIMA; FRANCHITO, 1993). Topography and bathymetry by National Center for Environmental Information (NCEI/NOAA).
3.2 PERIOD OF STUDY
The period studied covers the years 2009 to 2016. During this period, floods and water scarcity were recorded throughout the Brazilian northeast (MARENGO et al., 2017).
For example, the year of 2010 had a succession of intense rainfall which cause flood events in the eastern rivers of the Pernambuco and Alagoas States. The World Bank report and the government of Pernambuco (2010), registered a total of 67 damaged cities, 20 deaths, almost 30.000 homeless and an economic loss of approximate $ 1 billion. In that year, it was possible to observe positive SST anomalies in the WTSA which exceeded 1◦C in relation to the climatology between February and June, recorded in the Prediction and Research Moored Array in the Tropical Atlantic (PIRATA) buoy located at 8◦S;30◦W, as well as the SST data from the Advanced Very High-Resolution Radiometer (AVHRR) (http://oceanwatch.pifsc.noaa.gov).
In opposite, the year 2012 to 2016, were one of the driest periods in the Brazilian Northeast, registering droughts in the semiarid, as well as reaching part of the regions considered humid.
In this work we performed simulations for intense rainfalls (> 50 mm) of EWD across the years from 2009 to 2015. In total 16 events of intense rainfalls are simulated in years with high and low anomalies in SAWP SST.
3.3 EXPERIMENTS
3.3.1 Validations and 2010 Extreme Event
In the coupled experiments, the atmospheric model WRF provided atmospheric pressure values, zonal and meridional components of the wind at 10m, air temperature at 2m, fraction of clouds, precipitation, short and long wave radiation, as well heat and humidity fluxes. The ROMS model computes surface fluxes with the COARE bulk algorithm (FAIRALL et al., 1996). The oceanic model provides updated SST values for the atmospheric model (WARNER et al., 2010). The coupled interaction was configured to occur every 10 minutes of time step.
In the non-coupled model simulations, the initial conditions for SST was by the RTG-SST dataset. The update time-step it is 6-hours by interpolation of the initial data. In other words, the SST in WRF for these experiments don’t have interaction between air-sea.
The WRF model set-up and physics parameterizations were based on those used in the weather forecasting set-up applied by the APAC. The atmospheric model uses the Single-Moment 6-class scheme, which considers 6-class of hydrometeor setup, including water vapor, clouds and precipitation processes (HONG; LIM, 2006). Cumulus-type cloud parameterization was based on Kain-Fritsch (KAIN, 2004), as well the option of triggered moisture advection (kfeta-trigger) (MA; TAN, 2009) and sub-grid for the effects of convective and shallow clouds. The short and long wave radiation parameterizations used is from rapid radiative transfer model (RRTMG), which responsible for solving the surface energy balance (IACONO et al., 2008).
1. To verify the role of heat and moisture exchange over the TSA region, we disabled the sensible and latent heat fluxes at the water surface in the WRF-ROMS model (WRF-ROMS-E1) and the uncoupled WRF model (WRF-E1). The fluxes are solved only over the continent.
2. In this experiment, we applied the Mellor Yamada Nino Nakanishi (MYNN) parameter-ization for the resolution of fluxes and final analysis from GFS (FNL) data for initial conditions in the WRF-ROMS-E2 and WRF-E2 cases. This setup is used in the offi-cial meteorological service of Pernambuco State. The WRF-E2 will serve as the control simulation.
3. We used the same parameterizations from Experiment 2 but changed the input data to ERA-Interim. The simulations were named WRF-ROMS-E3 and WRF-E3. Another initial condition is used to verify how the heat fluxes respond to the changes in the meteorological fields and how they impact the rainfall simulations.
3.3.2 Test Case
It was performed two SST input data in contrasting years. The first condition used was with HYCOM SST from 10 of June 2010, with the normal conditions. The second with the
SST of HYCOM from 10 of June 2012, with colder SST. The oceanic currents, salinity and meteorological fields to input ROMS remained the same as 2010 conditions.
In the coupled experiments, the WRF provided Patm, V 10windand U 10wind, Tair, fraction of clouds, precipitation, short and long wave radiation, as well heat and humidity fluxes for June 2010. However, the oceanic model provides updated SST values for the atmospheric model every 10 minutes of time step with different SST conditions.
3.4 STATISTICS
Statistical validation was based on objective analysis of errors and concordance indices, as well as accuracy on set of atmospheric measurements compared to simulation results (HALLAK; FILHO, 2011). The tools described offer an alternative to statistical comparisons made in isolation at station points, which often penalise the results of numerical models (ANTHES et al., 1989).
The point-to-point correspondence between numerical simulation and observational measurements of the same variable provides a quantitative test to measure the model’s ability (or accuracy) to predict observed data. Thus, if a variable Xis and Xos, respectively, simulated and observed at a point in the domain, the fundamental amount for the study of errors is the difference (di) in time and space, expressed as:
di = Xis− Xos (3.1)
Basically, di = 0, perfect simulation; di or 0 imperfect simulation. Because it is a simple analysis, the estimate of the gross error does not show many types of errors that can contain a simulation, however derived from this analysis can be found several ways to verify such information.
The BIAS measures the tendency of the model to overestimate or underestimate the analysed variable in relation to the observed one. This trend, also called systematic mean error (ME) is defined as:
M E = BIAS = ¯d = 1 N N X i=1 di (3.2)
Bias does not convey information about individual errors and therefore can not be used as a measure of model accuracy. Moreover, while giving an idea of systematic bias or error, bias is affected by the fact that individual positive and negative errors of the same magnitude cancel out in the sum. The average of the absolute errors (Mean Absolute Error | d | or MAE) circumvents this problem. Because it is less affected by outliers, MAE is more accurate and robust as a measure of models’ ability to reproduce reality. The MAE is defined as:
